Hi Dmitry, thanks for looking at this. On 06/01/2016 07:59 PM, Dmitry Samersoff wrote:
Robbin,ProcessArgumentMatcher.java:61 What is the reason of keeping singlePid as a string. It might be better to convert it to Long at the argument processing time and then deal with Long.
VirtualMachineDescriptor returns pid as string, see ProcessArgumentMatcher.java:116 So I'll prefer to keep it as a string. Rest should be fixed, also contains fixes after input from Staffan (thanks). Inc: http://cr.openjdk.java.net/~rehn/8156537/02-01/webrev/ Full: http://cr.openjdk.java.net/~rehn/8156537/02/webrev/ /Robbin
Arguments.java: null seems more natural to me than "0" JInfo.java: 54 value of flag is not obvious to me, because optionCount do exactly the same. Update: it's the open question how we should proceed multiple -flag and -flags arguments. e.g. command line jinfo -flags -flag MinHeapFreeRatio=1 pidA pidA results different output for the same process. I didn't find any documentation specified it, So I would prefer to disable the combination like one above. At least for now. 74 missed space MonitoredVmUtil.java: 119 Could you change tmp to something more descriptive? Or ever better move the code guessing lastFileSparator to a separate function for better readability. -Dmitry On 2016-06-01 10:16, Robbin Ehn wrote:Hi all, There were some ripple effects, so to get everything working as intended the patch grow slightly. Webrev: http://cr.openjdk.java.net/~rehn/8156537/01/webrev/ Bug: https://bugs.openjdk.java.net/browse/JDK-8156537 Both a lot of manual testing, the new tests and jdk/test/sun/tools/,jdk/test/tools/ and some other which uses e.g. jcmd. Thanks! /Robbin
